Thursday, July 30, 2015

Final Presentation Outline

Disclaimer: This is only an outline or still a work in progress.  If you have any suggestions or comments please let me know!

Outline for Final Presentation:

Title Page: The use of Remote Sensing in Akagera National Park in order to analyze the impact of wildfires on savanna growth patterns by Anna Weldy

Background:
  • Akagera National Park background/history
  • Wildfires rage through the savanna ecosystem
  • Burns reduce plant fuel making fire in the next year less likely/drastic
  • Importance of fire in maintaining biodiversity
    • Burns reduce plant fuel making fire in the next year less likely/drastic
    • Ash from wildfire is rich in minerals, which attracts animals.

Objective:
  • To analyze, through remote sensing, how wildfires effect vegetation growth patterns over a span of four decades.
    • What can analysis of vegetation growth tell scientists about environment, climate..

Methods:
  • Landsat Data
    • TM was used (1984-present) because it has more bands and was therefore more practical for this analysis.
  • ENVI
  • Indices:
    • NDVI, SAVI/MSAVI, NBR, dNBR
      • Math/equations


Data/Images:
  • Four images from four decades before fires
    • SAVI/MSAVI computed for all four and compared
    • NDVI??
  • Use images after fires to explain changes over time
    • dNBR shows that a fire occurred and the severity of the burn.


Results/Conclusion:
  • What changes occurred over time, to what extent, how do we know...

Applications:
  • Fire management
  • Understanding change due to climate change/management

Thank you/Acknowledgments:


July 30: MSAVI and Bike trip

I started today by computing MSAVI for the 1980s image.  In order to do this, I had to calculate the slope of the NIR v. Red for a specific image.  Using this value and the equation NIR - (slope)(Red), I was able to compute the WDVI.  With the WDVI, I solved for L using another equation.  I used the L value to calculate the MSAVI.  In the end, MSAVI is calculated the same way that SAVI is, but SAVI assumes a the variable L to be constant at a value of 0.5.  MSAVI requires a bit more work because it relies on L to be calculated for the specific location and vegetation circumstances.

Then, at eleven, the interns and intern advisers went on a bike ride!  We rode to the University of Rochester and we visited Mount Hope Cemetery; we saw the graves of Frederick Douglas and Susan B. Anthony.  Then, we had a delicious lunch and rode back to RIT.  Even though it was EXTREMELY hot, it was a really fun trip (see picture below). 

After we got back and recovered from the extreme heat, I continued to work with SAVI for the image from the 1990s.  I won't be at work tomorrow, but I will post my outline tonight so that my fellow interns and advisers can see it and give me some constructive criticism.

Bike trip:

Wednesday, July 29, 2015

July 29

Today I worked on implementing the more difficult version of SAVI.  The SAVI I used yesterday uses infrared and red bands and an already predetermined constant; however, the version of SAVI which I used today needed additional calculations in order to find the constant.  I went through a number of equations and graphs and I am finally getting somewhere!  Today I also attended the weekly remote sensing meeting and all of the interns listened to a presentation about physics application to fire management.

July 28

Yesterday, I went through images from the 1980s and 1990s and I found a few images which will be used to show change over time.  However, when I got to images from 2004 and younger, the mask I tried to apply did not work.  It was rather strange, because the software recognized the mask as a mask band, but when it was applied the values of zero caused by the broken mirror in landsat 7 were not ignored.  I met with Dr. Vodacek to show him  the problem and he said that he would speak to someone today who is more familiar with the ENVI software.  For now, he wants me to work on applying SAVI to the older images and if we cannot get the mask to function, then he said I can just apply SAVI despite the anomalies caused by the broken mirror in landsat 7.

Monday, July 27, 2015

July 27: Week Four?!

Today I met with Dr. Vodacek in order to ask some questions on how to do analysis of fire burns over time.  He suggested that to start, I should choose four images from four decades (1980s-present) that were taken before fires and compare them to each other in order to note change over time. He also gave me a paper to read about SAVI and MSAVI so that I will know how to calculate this index in ENVI; he also helped me by explaining different strategies which will help me to check if I have built and applied a mask correctly.  I spent the remainder of the day reading the article and will start looking for the four images tomorrow.

Friday, July 24, 2015

July 24

Today I worked on cropping images and determining sections that I could use as subsets of data.  A smaller subset will be helpful because it will be more accurate and specific to the location.  We had our annual grill out and I also served as a test subject in the Color Science department.

Thursday, July 23, 2015

July 23: An apocalypse?!

At today's meeting we shared our progress and any problems we were having.  Then Mr. Callens took a group of interns over to see the Color Science department because we will serve as test subjects for a test going on there.  Then, all of the interns attended a presentation on space weather; it was so cool!  However, I never realized the severity of space weather, and apparently a severe enough event would cause humans to lose power for 10 years!  So now I am more than tempted to stock up on food and prepare for a near apocalyptic event because it could happen any day...

After the presentation, I kept sifting through images in order to distinguish fires and burn scars.  I got all the way up to 2004, which is when the Landsat 7 mirror broke.  So, I tried to work on building and applying a mask so that I could run an accurate NBR.  However, I do not think it worked, so I am planning on asking Dr. Vodacek for help tomorrow morning.  Today I also worked with the fires from 1984 and 1999 by running an NDVI and Tasseled Cap on both.

Wednesday, July 22, 2015

July 22

Today has been a very long day because the Mees trip ran rather late last night.  I started the morning by reading a paper which Dr. Vodacek sent me; it was about a census taken of the animals in Akagera National Park and therefore served as a good source for background.  I also did some additional research on the impact of fire on savanna ecosystems.  Then, Trevor and I attended the weekly Remote Sensing talk (Reagan came in later), which was actually a thesis defense: I hardly understood anything.

For the annual Wednesday talk, Dr. Messinger presented on a topic that I have become increasingly familiar with these past few weeks: remote sensing!  I saw parallels between his presentation and papers I have read; I learned a lot and gained a lot of clarity so the presentation was not only interesting but very applicable to me personally.

I spent my afternoon trying to pick out images of fires by both visually analyzing and by running a dNBR.  So far, I have found a fire in 1984 and a fire in 1999.  I do no know how many fires I am going to use, but I want to at least narrow down my pool of data so that I have a smaller number of images and do not feel so overwhelmed.  I am having a bit of a problem during layer stacking because I want to crop the images so they are all the same size and depicting the same location, and I thought I was doing it correctly.  However, when I go to run a dNBR in Band Math I am having to resubset the images again because they apparently aren't the same size; I am hoping to find a solution to this annoyance either now or tomorrow.

July 21: Stargazing

Yesterday was such an awesome day!  I came in for the first half of the day and spent a pretty relaxing morning doing some work.  First, I finished the mask that I had started on the day before.  Then, Dr. Vodacek emailed me back about my abstract; he gave me some really helpful pointers, saying that I needed to include more detail about the impact of fire in the Akagera savanna ecosystem and that I need to establish a period of time through which I will be working.  I spent the remainder of my day reading articles about the park and its savanna ecosystem.  I learned a lot, including that fires actually help to maintain biodiversity and to keep an ecosystem at its prime; I am going to read more in detail today about the actual ecology and biology that explains why fires are so vital to the ecosystem.

Later, at six o'clock, the interns, intern advisers, and the students in the graduate program boarded a bus to take to the Bristol Hills.  After a quick stop for dinner, we ventured into the remote areas of woods, wilderness, and no cell signal (gasp).  After about an hour bus drive, we climbed up a rocky path to the Mees observatory.  The night was absolutely beautiful and the telescope's size and innovation was crazy!  We arrived to the observatory when it was rather light out, so we used a smaller telescope to look at the moon.  Then, a group of interns, myself included, explored a hiking trail in order to kill some time for nightfall to set in.  Guided only by our smartphone flashlights and our supposedly good sense of direction, we actually did not get too lost and found our way back.  By the time we returned, the rest of the group had made it to the top of the observatory and it was dark enough to begin to see some stars.  I looked through the telescope a few times which was incredibly cool and unlike something I had ever done before.  In addition, simply with the naked eye, we spotted a satellite traveling across the sky and we could see numerous stars as the night grew older.  In the end, it was a really incredible trip because I had never been to such an observatory and have never really stargazed; and, it was also a great way to get to bond and spend time with all of the interns.  All and all, a great day.

Monday, July 20, 2015

July 20: Week Three!

I spent today by picking up where I left off on Friday: creating masks in ENVI.  I needed to create a mask to cover/ignore all of the distortions caused by the broken mirror in Landsat eight.  Basically, the broken mirror caused lines to form on the images, and instead of actually collecting data, these lines all show zero as the data.  Since the data that would be underneath these lines does not have a value of zero, having incorrect data when running indices would effect averages, means, and other calculations which would throw off the entire project.  So, I spent the morning surfing the web and the ENVI help page in order to figure out how to apply a mask.  I thought that I could simply tell the software to ignore data values of zero (the obscuring lines), and then a mask would be created quite simply.  Unfortunately, the software only allowed me to input a range of numbers, which was not helpful considering that I only wanted to mask one numerical value.  After a morning with very little luck, I went to Dr. Vodacek to ask him what I should do.  He reasoned that creating and saving a region of interest (RIO) would be the best way to do it.  To create a region of interest, I have to click every line individually and then save the RIOs as one file.  I wish that there were a way to just compute a numerical value to ignore, because clicking each line can be a bit painstaking and takes a lot more time.  However, at least now I have a way to mask inaccurate numbers which gives me access to many more images that would otherwise be obscured.

Friday, July 17, 2015

July 17: Masks and the picnic table incident

It is crazy to think that we are already a third of the way through the internship; I am kind of sad but really excited for the next four weeks!  At today's intern meeting, I presented my abstract to all of the other interns and I listened to the other abstracts.  I received some feedback which I applied to my abstract (see yesterday's post).  After the meeting, I researched a bit more about SAVI and a few other indices (EVI, SATVI, LAI, NPP, etc).  I do not know if I will need or use these indices in my project, but now I have some background on all of them to the point where I would feel comfortable applying them.  In addition, I did some brainstorming on the different factors that influence both vegetation and fire severity; I wanted to come with a few ideas of different variable combinations that I could graph that might show vegetative growth and patterns.  I also did some research about Akagera and I viewed the park both on google images and google maps in order to familiarize myself with it; it is absolutely stunning!

At noon, myself and my fellow interns grilled hot dogs and kebabs and played some strange variation of volleyball (don't ask).  Also, nearly five interns died due to a picnic table and some soda, but that's another story (and no one was actually in any physical danger.  Well, not really.....). After lunch, I started applying the NBR to two images from 2004, but there were some strange lines that distorted the image.  Dr. Vodacek explained that these lines and distortions were caused by a broken mirror in Landsat 7.  In order to fix the distortions and have the images still be useful, I have to create a mask which will, you guessed it, "mask" the distortions so that they do not alter or disrupt calculations.  I started reading a little bit about masks, but they are a bit confusing because there are so many options on how to mask certain sections.  I hope that I can figure out masks by Monday so that the distorted images can be used for my project.

Thursday, July 16, 2015

July 16: Donuts and Final Presentation Abstract

Today started out like every day should: with donuts!  Following our intern meeting, I used ENVI in order to find the NBR for images from June and July of 1984.  To find the NBR, I actually used the NDVI tool, but I inputted a different band number and it gave me a complete different index; I thought that was pretty neat.  I also used this cool tool called band math in order to subtract the pre and post fire NBRs and to find the difference.  In addition, I did some research on other indices that I read about in one of the papers Dr. Vodacek gave me last week.  One that really stood out to me was the SAVI, or soil-adjusted vegetation index. This index is used in areas where vegetation cover is low, and it works similarly to the NDVI.  The main difference is that the NDVI is measured using plant reflection, whereas the SAVI is a measured using soil reflection.  Because the area of Akagera National Park I am interested is a savanna, SAVI will probably become very useful because there is less plant coverage in a savanna then there is in say, a forest.  I finished up the day by drafting up my abstract for my final presentation (see below); all of us interns are presenting our abstracts at our daily meeting tomorrow morning.

Abstract for Final Presentation

Forest fires rage throughout various ecosystems in the world, causing intense damage to the areas through which they go.  Analysis of forest fire and its effect on an ecosystem can aid biologists, ecologists, and other professionals in further understanding that particular ecosystem; it can also aid in solving and preventing problems.The objective for the six week internship is to discover, using remote sensing, how wildfires and the fire scars they leave impact vegetation growth patterns.   Specifically, this project is looking at the Akagera National Park located in northeastern Rwanda.  Through the use of Landsat satellite images and various indices, one can identify both burn scars and the severity of the burn scars by using the NBR, or normalized burn ratio, which helps to quantify severity of the burns.  Likewise, a variety of other indices can help to quantify vegetative health and growth.  Through the use of these various indices as well as other tools, one can observe and analyze the images before and after a fire in order to note any changing vegetative growth patterns over time.

Wednesday, July 15, 2015

July 15

Today I started by reading a few papers that Dr. Vodacek gave me on the area of my research project: using remote sensing in order to identify fire burns/scars.  In order to complete this task, most researchers use the NBR, which stands for normalized burn ratio.  By taking the difference of the near infrared band 4 and the short wave infrared band 7 and dividing this value by the sum of those same bands, one gets the NBR for the particular region in which they are interested.  If one then finds the NBR for a particular region before and after a fire comes through, the change in NBR can be solved by finding the difference of the pre and post fire NBR values.  On a scale, the NBR can show the severity of the fire and therefore aid in the analysis of the impact on vegetation.  In addition, today I attended the annual Remote Sensing Meeting and all of the interns sat in on a presentation by Mr. Pow about the freshman imaging project that incoming freshman at RIT take part in.  My goal for tomorrow is to apply what I learned today about the NBR to images with noticeable fire scars and burns.

Tuesday, July 14, 2015

July 14

I spent today finally looking at the images that I am analyzing!  I looked at all of the images which I downloaded from the Landsat 1-3 MSS device; these images are from the early 1980s and therefore have fewer bands than the Landsat 8 device.  I messed around with the NDVI, which stands for normalized difference vegetation index; it is basically an index which measures the photosynthetic activity of plants.  I also viewed bands which had wave lengths in the near infrared region because healthy plants reflect near infrared and less healthy plants absorb it.  What I don't understand is how I can identify burn scars because decrease in vegetation can be caused by a number of factors other than wildfire, including drought, less foliage cover, or simply because it is the wrong season for that particular plant. My confusion lies in solving this problem, as well as understanding what other tools can help me with my analysis and how to quantify and qualify changes in vegetative patterns as a result from fire burns and scars.  However, I have sent an email to Dr. Vodacek so I should have some answers in the morning!  Today I was also a test subject for Jason and John, who are the two interns working in the eye tracking lab, and I sat in on a thesis defense in the afternoon; both were very cool!

Monday, July 13, 2015

July 13: Beginning of Week Two

Today was a pretty uneventful day because I spent the day downloading satellite images of the Akagera National Park in Rwanda which is the location in which I am studying fire burns and vegetation patterns with the help of remote sensing.  The files are all very big and because I was downloading all useful images from thirty plus years ago, it took quite a long time.  Thankfully, I have just a few more images to download so I hope I can really start analyzing and attacking my project this week!

Friday, July 10, 2015

July 10: End of Week One

Today was a really fun day.  After the intern meeting in the morning, I relabeled the different bands of the image I was observing in Rwanda based on wave length.  This way, when I look at different band combinations, I will be able to use the wavelength in order to know if what I am looking at is infrared, color, thermal, etc. Afterwards, the interns and staff of the CIS building got together for the annual Friday grill out and volleyball game.  It was so much fun and slightly intense.  Afterward, I spent the remainder of the day downloading a ton of images of the Akagera National Park, some dating back to the early 1970s and 1980s, so that I can begin to analyze them next week!

Thursday, July 9, 2015

July 9

Today started with the usual intern meeting in the Reading Room.  This morning, however, we went around discussing problems and successes we are currently having in the different work we are doing; all of the other intern projects sound really interesting and cool.  Following the meeting, I continued to try to work to figure out how to save an image with multiple bands as one file on the newer edition of the ENVI software.  After awhile, Dr. Vodacek came to the conclusion that it would be easier to use the older version of ENVI as it is more straight forward and useful for my project.  After making that switch, it was a lot easier to save and access all of the bands in one place.  I messed around with different band combinations and then took my lunch break.  During my break, I met with Nate and Angela, the two interns who are working in the Optics, and they showed me the project they are working on.  In the afternoon, I worked a bit more with different band combinations in ENVI and I spent the remainder of the day reading papers that Dr. Vodacek gave me so that I might become more familiar with the work that I am doing.

Wednesday, July 8, 2015

July 8

Day three!  Following our daily morning meeting, I began to work with the software, which I downloaded yesterday, called ENVI.  Basically, it allows one to process images and extrapolate different bands in order to analyze and examine the images more clearly and carefully.  At 10 am, I attended the weekly Remote Sensing Lab meeting.  For an hour or so, the entire faculty and staff of the Remote Sensing department get together and one of the faculty makes a presentation on work which they are doing in this area.  Today, Dr. Vodacek, who is also my advisor, presented his work in tracking moving vehicles; his work combines both real images and the anticipated kinematics of a vehicle in order to track where the vehicle drives. While the presentation was very upper level and I am not familiar with all of the terminology and technology used, I felt that I was able to grasp the basic concept and I found the ideas and concepts presented to be really cool and innovative.  It was also really neat to see and listen in on research that is ongoing and to hear colleagues discuss ideas.

 Following the meeting, I continued to work to access satellite images and started to familiarize myself with ENVI.  Then, at noon, some of the interns, myself included, met with undergraduate students who are working here over the summer in order to listen in on a presentation about historical document imagining.  Dr. Easton, who works in the historical document imaging lab, explained how he used different waves of light in order to uncover texts that would otherwise be unseen to the human eye, which was really cool.  Afterwards, I went back to working on ENVI and finished the day by accessing the specific location which I will be looking at and analyzing for the next six weeks; this location is the Akagera National Park located in Rwanda.

July 7

Yesterday started with a walk across campus to the rock climbing facility called Red Barn.  At the Red Barn, interns and intern leaders engaged in a series of team bonding and building activities.  It was really fun to work with my fellow interns in order to solve problems and to accomplish the tasks that were set before us.  While we were not necessarily successful with each task presented to us, we all learned a lot about skills needed to problem solve and we got to know each other a bit better.  In the afternoon I met with Dr. Vodacek and we spent the remainder of the day downloading ENVI, which is a software that allows us to view and analyze remotely sensed data.  After a few frustrating technical issues, we finally fully installed the software; I am really excited to learn how to use it and to hopefully become emerged in data analysis starting within the next few days!

Tuesday, July 7, 2015

July 6: The First Day

Yesterday, July 6th, was a really exciting and fun day as it was the start of the internship.  I met with my fellow interns and our internship leaders in the morning and we spent the first half of the day touring the CIS building where we will spend the next six weeks working.  In addition, we performed a team building activity which involved us completing a list of tasks given to us by our internship leaders.  After lunch, we met with our individual advisers; my research adviser is Dr. Tony Vodacek in the Remote Sensing lab.  My project consists of viewing different satellite images of the Earth at various different wave lengths.  The specific location I am looking at is a park in Rwanda; I will analyze burn scars from forest fires and see how these burn scars directly relate to changing patterns of vegetation growth.  While yesterday was just an overview of the project, there will be more detailed updates to come.  I am really looking forward to the next six weeks with the Remote Sensing lab team!